Re: subsref

["Jiro Doke"]

Gucky wrote:
>
>
> Hi!
>
> I have a problem with overloading SUBSREF
>
> I have an object A with e.g. two members like A.name and A.age, if
> I
> have vector of this objects my program make this vector to an
> object
> of the type A. But if I want to call now A(2) it is (of course) not
> possible, because A is one object. So I thought I overload subsref
> in
> a way, that this call a routine in subsref, which extract all the
> data of A, which is part of the previous second element.
> But the main problem is, that A(2) and e.g. A.name(2) are calling
> both the same ( ) operator, is there a way to find out if there is
> a
> "." before or after the ()?
> Or have someone another idea to handle that problem?
>
> Thank you for helping!

Hi.
I'm not exactly sure how you have to object setup, but in the
subsref.m, you have

function B = subsref(A,S)

right? S includes all of the indeces called. So if you call
A.name(2), S will have 2 structures, one for '.' and one for (2). So
maybe you can code it so that it works recurrsively. You would only
process the first S structure, and recurrsively call subsref for the
remaining S. Then it would call the appropriate function depending on
the class. This probably doesn't make sense, so here's a sample
subsref.m:

function w = subsref(w,index)

switch index(1).type
  case '()'
    i = index(1).subs{1};
    if i <= length(fieldnames(w))
      fn = fieldnames(w);
      % call i-th field
      w = w.(fn{i});
    end
  case '.'
    if any(strcmp(index(1).subs,fieldnames(w)))
      w = w.(index(1).subs);
    else
      error(sprintf('Invalid field name: %s\n',index(1).subs))
    end
  case '{}'
    error('Cell array indexing not supported by this object')
end

% if more in index, recurrsively call subsref for the remaining
if length(index) > 1
  w = subsref(w,index(2:end));
end

jiro